Emotional processes appear to generally decline with age, particularly responsiveness to negative events and stimuli. Carstensen's (e.g., 1992) socioemotional selectivity theory (SST) is an account of this effect that enjoys wide empirical support. However, some studies find that older adults do not differ from the young on measures of perceptual speed (Mather &Knight, in press) or attentional capture (Larsen &Beaudreau, 2005) when the target objects are affective (particularly threat or fear) stimuli. How can older adults differ from younger in their responsiveness to emotional stimuli, yet not differ in their perceptual processing of, or attention to, those stimuli? In this application we propose the answer lies in applying a dual-process model of emotion to life-span questions about emotional change. According to this model, one emotion process involves relatively fast, early-stage, pre-attentive and automatic processing of the emotional significance of a stimulus. A second process involves a more controlled, effortful or executive function later-stage evaluation of personal meaning of that stimulus. We hypothesize that the early stage process - the automatic component - remains intact with age and is preserved well into later life and may even be resistant to the onslaught of Alzheimer's disease. The more controlled late-stage process, however, is where age differences are most likely to be found. Indeed, Carstensen's SST may apply mainly to the more controlled, later-stage aspects of emotional responding. In other words, older subjects register and initially process emotional information the same as younger, but then they differ in terms of how they regulate that initial responsiveness. Three lines of investigation are proposed to test these hypotheses. One series of studies will use laboratory tasks that assess early-stage processing of emotional stimuli (e.g., emotion priming tasks, affective Simon and Stroop tasks, pre-attentive dot probe tasks, incidental learning task, and classical conditioning to negative stimuli) to examine whether older adults demonstrate similar negative affect biases routinely found with younger adults. The second line of investigation will employ the process dissociation procedure (Jacoby, 1991) to separate automatic and controlled processes in emotional (evaluative responding) tasks. Here we propose that older adults will exhibit the same automatic bias component found with younger adults, but will differ in the controlled component of their evaluative responses. A 2-year longitudinal study is also proposed to exam the automatic component of emotion in patients suffering from early-stage Alzheimer's disease. We predict that, even in the face of cognitive decline, the automatic component of emotional responding will remain preserved in Alzheimer's patients. Taken together these studies can resolve a major inconsistency in the emotion and aging literature and lead to a deeper understanding of which emotional processes do and do not change with normal aging and with Alzheimer's disease.